On Wed, 2010-05-05 at 17:55 +0000, Carl Eugen Hoyos wrote: > Uoti Urpala <uoti.urpala <at> pp1.inet.fi> writes: > > > There are several unfixed VDPAU-related issues in svn btw > > You mean like "performance of de-interlacing is better with svn"? I mean the various bugs, limitations and performance disadvantages in svn. Are you seriously trying to give the impression that svn would _not_ be significantly worse? IMO that's quite silly given how obvious the differences are. There is one specific case with software decoding, field-separating deinterlace and particular display refresh rate/playback speed ratios where git spends more time waiting in the VO (when using hardware decoding git is faster; at higher display refresh rates the effect doesn't occur, and at lower refresh rates svn fails instead due to its flip limitation). I think I can improve that case pretty easily (the details of the current code are partially due to problems with earlier NVIDIA drivers, which have been fixed in later versions). And on the other hand in svn _all_ field deinterlace cases have at least timing problems and limitations such as framestep skipping the added fields.
